Distributed Event Processing

Distributed Event Processing : 

Distributed Event Processing Institut für Parallele und Verteilte Systeme Universität Stuttgart

What are event-based applications? : 

Event Processing What are event-based applications? © 2006-2010 1 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Examples: Our contribution: Event service that provides advanced event processing to applications Event-based applications: Send and react on events Often distributed Support large number of participants Event: An occurrence E.g. “a car moves” Complex events reflect a pattern of events E.g. “accident on a highway” Peer-to-Peer multi-player online role games Augmented-Reality Services Vehicular networks

Event service for the SpoVNet project : 

Event communication Content-based forwarding Content-based communication requires no addressing Matches individual interest with message content Complex event processing In-network situation detection Processes data while in transit Creates high-value information from low-level data © 2006-2010 Universität Stuttgart, Institut für Parallele und Verteilte Systeme 2 SpoVNet Event Service Event service for the SpoVNet project Designed as self-organizing overlays on SpoVNet nodes

Challenges for distributed Event Processing : 

SpoVNet Event Service Challenges for distributed Event Processing © 2006-2010 3 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Expressiveness Situation specification Event filters Heterogeneity Devices, communication Systems, models Self-organization Migration Adaptation Security Confidentiality Authentication, authorization

Our contributions to Event Processing (I) : 

SpoVNet Event Service Our contributions to Event Processing (I) © 2006-2010 4 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Self-organization Subscription clustering Based on subscription model Based on event history Incorporation of application semantics into forwarding Efficient forwarding trees Adapts to dynamic applications Awareness of heterogeneity Resource-aware placement of service functionality Access of Cross-layer information Specification of application requirements using a Restriction Graph

Our contributions to Event Processing (II) : 

SpoVNet Event Service Our contributions to Event Processing (II) © 2006-2010 5 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Quality of Service Guarantee of delay bounds Saving bandwidth Minimizing “false positives” Measuring situation detection confidence Security Asymmetric encryption Achieves weak confidentiality Attribute-based access control Keys valid for epochs Two subscription models Attribute/value pairs Spatial indexing QoS specification allowed in subscriptions Expressiveness Flexible & exact situation description language Detection algorithm with full “selection/consumption” policy support

Under Development : 

SpoVNet Event Service Under Development © 2006-2010 6 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Modularization Plug your own event service together Shared overlay functionality use with other services Quality-aware correlation detection Application-dependent specification of detection quality Integration of sensor networks Situation detection on sensor data

Questions / Feedback? : 

Contact Information Questions / Feedback? © 2006-2010 7 Universität Stuttgart, Institut für Parallele und Verteilte Systeme Get information, read papers, download software: Contact us: Visit us: Situation detection: Gerald G. Koch Project division leader Dr. Boris Koldehofe IPVS – Institute for Distributed and Parallel Systems University of Stuttgart Universitätsstr. 38 D-70568 Stuttgart http://www.ipvs.uni-stuttgart.de/abteilungen/vs/start/en www.spovnet.de Event communication: M. Adnan Tariq Head of Institute Prof. Dr. Kurt Rothermel